|
Recovered from the surface of the three-dimensional set of sampling points , the geometric model of the surface is called surface reconstruction . Surface Reconstruction many research areas , such as reverse engineering , an important issue in the medical image visualization , but also a hot research . This paper studies the unordered point cloud 3D reconstruction , the research point cloud reconstruction theory based on the following questions in-depth discussion and analysis : First , in order to ensure the redevelopment effect , remove the problem of noise in the point cloud the study , the first analysis of the existing point cloud denoising algorithm , and then discussing how to establish topological relations of unordered point cloud based on disordered point cloud denoising denoising algorithm based on kd tree . Secondly, this paper, a theoretical proof of the Power Crust algorithm based on the Voronoi diagram . The algorithm , regardless of the uniform sampling , or non- uniform sampling unordered point cloud , have a better reconstruction results . But the point cloud reconstruction of large magnitude too slow . To solve this problem , this paper presents a Power Crust algorithm based point cloud reconstruction algorithm . The original sampling points before surface reconstruction , according to the local features of the sampling point scale inhomogeneity down-sampling , to meet the the reconstruction conditions the premise assurance sampling points under reduce computing sampling point , thus reducing the amount of computation of the reconstruction algorithm , greatly improve redevelopment speed . In order for the model after reconstruction smoother surface , close to the original model , this article describes the Laplace smoothing algorithm . Finally, the combination of VC6.0 and VTK Power Crust algorithm of this paper, the fast and the zero set method of Hoppe developed a point cloud reconstruction system , grid smoothing , according to the characteristics of the number of point clouds and sampling Select a different reconstruction methods .
|